Order in Chaos Intra Party Coordination in Open List PR Systems José Antonio Cheibub ([email protected]) Gisela Sin ([email protected]) University of Illinois at Urbana-Champaign Presented at the 72nd annual Midwest Political Science Association Conference, April 3-6, 2014, Palmer House Hilton, Chicago, IL. Note: This paper is incomplete and represents work in progress. We would appreciate any comments and suggestions you may have for further development. We thank Lorena Barberia, Argelina Figueiredo, Christopher Grady, Will Heller, Fernando Limongi, and Charla Waeiss for comments on an earlier version, and Matthew Rowen for excellent research assistance. 1. Introduction The purpose of this paper is to study intra-party coordination in open list proportional representation systems. Specifically, we show that political parties play a central role in composing party lists and that, in so doing, they successfully prevent the eruption of all-out competition among co-partisans. We provide evidence of intra-party coordination for contemporary BraZil, which is, arguably, the worst case for finding such evidence. 2. List Proportional Representation Electoral Systems One of the most fundamental distinctions made when we think about electoral systems is that between majoritarian and proportional principles. As Powell (2000) has argued, these principles inform two different conceptions of elections and representation, which generate distinct ways in which voters, politicians and governments relate to each other. Yet, proportional representation systems (PR) are not all alike, and not simply because they use different formulas for translating votes into seats. One central distinction is that which separates list PR systems that allow for preference vote and those that do not.1 In the latter voters are presented with and vote for ranked and closed lists of candidates. If a list gets n seats, the first n candidates in that list are the ones to whom the seats are assigned. In the former, voters are presented with lists that may or may not be ranked, but they are never closed: voters are allowed to express a preference for candidates within the list and, in this way, affect which individual candidate will be elected. In spite of considerable variation in details, preferential list-PR systems have at least two things in common. First, a collective entity (a party or candidate list) is recognized as the primary recipient of seats. Before seats are allocated to individual candidates in some relation to the number of votes they obtained, they are allocated to the list as a whole. Second, voters are allowed to have a say in allocating seats to candidates within the party once the number of seats for the party has been determined. In some cases this only means the ability to change the order of candidates originally proposed by the party; in others, it means collectively ordering an entire list of unordered candidates. Open-list PR systems, our focus in this paper, belong to the last category; they are an extreme version of preferential list-PR systems in the sense that the rank of candidate within the list is entirely determined by voters: if a list gets n seats, the n most voted candidates are the ones to whom the seats are assigned. Table 1 lists the countries that currently adopt preferential list- PR as the rule for electing their legislatures. 1 This is a distinction that applies to list PR systems. There are other electoral systems that produce 1 Table 1 Preferential PR Systems in the World (bold = Open List) WESTERN EUROPE EASTERN EUROPE LATIN AMERICA ASIA Austria Liechtenstein Bosnia Herzegovinia Brazil Indonesia Belgium Luxembourg Estonia Chile Japan Cyprus Netherlands Latvia Colombia Sri Lanka Denmark Norway Poland Peru Finland Sweden Slovakia Greece Switzerland Slovenia These two features of preferential list-PR are considered decisive for the kind of electoral strategy candidates pursue. They imply that OLPR systems are characteriZed by double competition: parties compete with one another to maximiZe the number of seats they will get; candidates, in turn, compete with members of their own parties for the votes that will guarantee that they will be the ones to be elected. For this reason successful candidates cannot simply rely on party labels: success in elections requires that, in addition to, or at the expense of, their parties, they themselves be recogniZed and chosen by voters. Thus, in order to be successful in OLPR electoral systems, candidates must develop a strong direct link with their districts, which they do by providing particularized goods. Given that districts are multimember, however, the best way to do so is to create de facto single-member constituencies where the link between the politician and voters can be nurtured and sustained without the constant need to distinguish oneself from intra- and inter-party competitors. In this way, politicians operating in open-list PR systems are presumed to behave much like politicians under single-member plurality elections, that is, they focus their actions in office, or the promises they make in order to get into it, on geographically targeted policies and on the provision of services to their constituents (Cain, Ferejohn and Fiorina 1987, Ames 2000). This strategy contrasts with the one pursued by politicians operating in closed-list PR systems, where the incentives for cultivating personal reputations are low and, as a consequence, broad, partisan considerations take a central place in the politicians’ behavior and promises. The corollary is that in open-list PR systems candidates have strong incentives to cultivate personal reputations, injecting the system with a high degree of personalism and with weak political parties. Because politicians must compete with candidates from other parties and with candidates from their own party, competition is individualiZed and political parties have little or no role to play in organizing the electoral contest. As recent studies have argued, this feature of preferential list PR, but particularly of OLPR, is at the root of a number of important outcomes. For example, Shugart (1999) argues that collective goods at the national level are more easily provided in “party-centered” elections (of which closed-list PR) 2 than in “candidate-centered” elections (of which preferential PR). Milesi-Ferretti, Perotti and Rostagno (2002) and Nielson (2003) argue that public policies are more narrowly targeted when legislatures are elected through OLPR rules. For Samuels (2004), “nationaliZing” electoral systems, defined as systems employing closed-list PR, in moderately large districts enhance accountability (as measured by the change in the share of votes obtained by the incumbent party between two elections). Nicolau (2006), in turn, argues that accountability suffers in OLPR systems because of the way votes are counter and seats distributed in these systems. Carey and Shugart (1995) hypothesiZe that systems with “party-centered” rules will have lower levels of corruption, voters will vote on the basis of broad policy options, will have a higher overall degree of economic liberalism in trade and industrial policy, and higher levels of legislative party cohesion. Jones and Navia (1999) find that gender imbalance is greater in open- than in closed-list PR systems. Kunicova and Rose-Ackerman (2003) and Persson, Tabellini and Trebbi (2003) find that corruption is lower under open- than under closed-list PR. Chang (2005) and Golden (2003), on the other, hand, argue that much of Italy’s pathologies, including its inordinate level of corruption, can be directly traced to the incentives generated by the OLPR system used until 1994. Gingerich (2013), however, argues that what differs across these systems is the type, not the level, of corruption. Finally, Hallerber and Marier (2004) associate OLPR rules with higher budget deficits and Hicken and Simmons (2008) with inefficient public spending. 3. Coordination in Open-List PR Systems: The Current View We have a good understanding of electoral coordination in majoritarian systems. Electoral coordination – a collective phenomenon – will happen when voters and other political agents individually decide to behave strategically. Strategic behavior in elections, in turn, means that actors take actions “because of their perceived impact on the final outcome of the election, rather than because of any intrinsic value they may have” (Cox 1999: 149). As Cox (1997) has demonstrated, given a majoritarian electoral system, this kind of coordination will happen when some specific conditions are in place, including instrumentally rational voters and shared information about candidate placement Under these conditions, the result of successful coordination is the reduction in the number of candidates or parties competing in any given electoral district to M + 1, where M is the magnitude of the electoral district. In PR systems, however, M + 1 offers no more than an upper bound for the number of candidates or lists that will compete, a number that is of little, if any, relevance given districts above a certain magnitude (about 5 according to the existing literature; see Cox and Shugart 1996 and Lago 2011). PR formulas applied in relatively large districts are considered to be permissive in the sense that institutional barriers to enter competition are low. The limit to the number of parties and candidates competing for votes in large district-PR systems is sociological; it is a function of the diversity of interests that find collective 3 expression in the district.2 Thus, once organized into parties, interests will get seats more or less in proportion to the number of votes they obtain. Political parties are crucial for the stability of large district-PR systems. Given the system’s permissiveness, it is vulnerable to a high degree of fluidity. Party organiZations, however, act to prevent fluidity and to prevent conflicts that may lead to defections and the emergence of new parties. Thus, whatever order there is in large-district PR systems, it results from the activity of political parties themselves.